Latest Open Artifacts: Motif-3, GLM-5.3, Hy4-Preview, and Open Licensing Trends
The open-weight AI ecosystem has expanded with notable releases including Motif-3 (a 314B parameter MoE model), Zhipu AI's GLM-5.3 reasoning model, and Tencent's 770B parameter Hy4-preview. Alongside these new model weights, the ecosystem is actively refining open AI model licenses. The release of massive open-weight Mixture-of-Experts models gives developers and researchers powerful alternatives to proprietary LLMs. Clarifying and evolving open model licensing is equally vital for establishing legal boundaries for commercial deployment and community collaboration. Tencent's Hy4-preview features 770 billion total parameters with 49 billion active per token across 78 layers and supports a 1M+ token context window, while Motif-3 utilizes 314 billion parameters with 13.2 billion active per token. Zhipu AI's GLM-5.3 targets complex software engineering and long-horizon agent tasks with improved token efficiency and a 1M token context window.
## BACKGROUND
Open-weight models allow developers to download and execute neural network parameter weights locally or on custom infrastructure, unlike closed APIs. Mixture-of-Experts (MoE) architectures enable models to scale up total parameters while maintaining manageable computational costs by routing each token to only a small subset of experts during inference.